Cloud Chamber 76
August 1997

My profoundly philosophical thought of the month, confided to a few admiring acolytes at the Jubilee, is: 'It's too bloody hot.' Consequent brain death led to the now traditional BSFA meeting date cock-up in Ansible: this month's is of course 27 August and not (er, um, shuffle, evasive sideways looks) 23 July.

Maths Commonplace Book. Arthur D.Hlavaty on Lancelot Hogben's Mathematics for the Million (1936): 'What Hogben seemed to be saying to his target audience was, "Because you are not white, male, and upper class, you have been told you can't do math. This is untrue, and this book will help you do it." Now we are hearing more and more, "Because you are not white, male, and upper class, you can't do capitalist oppressor things like math. This is a good thing, and this book will help you feel better about it."' • 'Roses are red / Violets are blue / Fermat is dead / But his theorem's true.' (Anon, Usenet). 'My improved value of Pi does away with all this waste. I'm not allowed to publish the new value – it is of course classified – but here's a hint: From now on, all circles are going to be a whole lot rounder.' (John Sladek, 'How to Make Major Scientific Discoveries at Home in Your Spare Time') • 'Old Macdonald had a farm, 2.7182818285, root -1, 2.7182818285, root -1, 0 ...' (Trad maths graffito). • 'It enables a numskull to memorize a quantity of numerals.' (Mnemonic for famous constant cited above)

Another Urban Myth? Received under the title 'The Best Heckle Ever?' ... At a recent Sacramento PC User's Group meeting, a company was demonstrating its latest speech-recognition software. A representative from the company was just about ready to start the demonstration and asked everyone in the room to quiet down. Just then someone at the back yelled, 'Format C: Return.' And someone else added: 'Y, Return.'

Unfortunately (the story goes), the software worked....

Read in Reading

Douglas Hofstadter and others, Fluid Concepts and Creative Analogies: Computer Models of the Fundamental Mechanisms of Thought (1995; 1997 UK) ... a heavy-sounding title, but I loved Hofstader's Gödel, Escher, Bach and Metamagical Themas too much to hesitate over this one. The previous books foreshadowed his basic AI philosophy, which is that the ability to see patterns, devise analogies and 'slip' from analogy to analogy lies at the heart of human thought and creativity. According to Hofstadter, much vaunted AI work (examples of which he gleefully dissects) has engaged with problems too complex, like recognition of human faces or generation of meaningful prose, for the extent of its failure to be apparent – e.g. human wishful thinking too easily conjures up a personality behind utterly uncreative computer-produced prose ('the Eliza effect'). Hofstadter's 'Fluid Analogies Research Group' has been tackling highly simplified analogical problems with whimsical project/program names like Seek-Whence (looking for underlying patterns in number sequences), Copycat (analogies between letter strings: what is to xyz as abd is to abc, when the alphabet may not be considered as cyclical?), Tabletop (positional analogies jokily using plates and cutlery) and Letter Spirit (a program to recognize the underlying 'spirit' of a much simplified font, and generate more letters in the same 'style'). The programs are described in outline – a bit repetitious since they all have a similar structure, but the structure is fascinating and plausible (well, to me) as a model of low-level thought, with 'ideas' being processed at random but with those guesstimated as more 'important' or 'urgent' being assigned better chances, and with feedback that repeatedly recalculates this importance or urgency. What's interesting about the results isn't that Copycat always gets the 'right' answer to 'If abc becomes abd, what does xyz become?' In fact there's no one right answer, though xya is ruled out because z can't be assumed to have a successor. Most often the program comes up with the 'dumb' solution chosen by many people, abd (assumed rule: 'replace rightmost letter with d'), but does not rate it highly for elegance; sometimes it hits on the neater wyz (resulting from two creative inversions) and does give it a high elegance rating; the in-between answer yyz is occasionally chosen but is regarded by Copycat as even less elegant than abd. Of course the actual problems being tackled here are of mind-wrenching unimportance ... but much more subtle patterning can be encoded in those tiny letter-string analogies, and Copycat and its siblings manage some other impressive 'insights'. Hofstadter's team could be on the trail of important things ... although I think it'll be a little while before any of these programs announces, 'Yes, now there is a God!'

Also ... A.K.Dewdney, Yes, We Have No Neutrons: An Eye-Opening Tour through the Twists and Turns of Bad Science (1997), a rather lightweight tour of science gone bad – mistaken or pathological, rather than fraudulent. Alas, most of the examples have been extensively covered in past sceptical texts: cold fusion, IQ tests, N-rays, the usual pantheon. It comes to life when the material is newer (e.g. the shoddy science of that weird Biosphere 2 project) and/or lies in Dewdney's own patch of computer science (hence a swingeing attack on the excessive claims made for neural networks as the Universal Computing Panacea). • L.E.Modesitt, The Death of Chaos (1996): yet another 'Recluce' fantasy full of bland prose and tiresome onomatopoeia (Mark and I agreed that the most disorienting sound effect is the Wheee-eeee which LEM fondly believes to be the noise made by horses). Further bad guys – 'chaos wizards' – bite the dust, but it all feels a bit futile since the supply is endless: in FE terms, the series deviates from the pattern of fulfilled fantasy by not permitting HEALING.... • Joan Aiken, The Cockatrice Boys (1996), very odd YA fantasy about a future Britain overrun with monsters, from Basilisks to Snarks: light-hearted telling and a high death toll. • Clifford Ross & Karen Wilkin, The World of Edward Gorey (1996): two OK essays on Gorey, and something like 200 of the great man's sinister and tasty illustrations, including unpublished material. Perhaps too much familiar material here if (like me) you already collect Gorey, but it's a fine sampler. Next, go and buy Amphigorey.

On the Heap ... Douglas Hofstadter, Le Ton beau de Marot (1997), another multi-stranded blockbuster almost as big and dense as Gödel, Escher, Bach but probably more accessible to those who suffer from Fear Of Maths: it's about languages and translation (with wide digressions), but I haven't got very far yet. • Greg Egan, Diaspora (1997) – for SFX review.

Mailing 55. You were all just wonderful, but it's still too bloody hot; and the copier's died, so I need to stop at one laser-printed side. That Mars chat-show: spies tell me that Clive Anderson wanted to give the guests free rein, but was put under strict orders to keep breaking in with relentless one-liners from the autocue. [...]